ROBOHOW.COG · Web-enabled and Experience-based Cognitive Robots that Learn Complex Everyday Manipulation Tasks

Enabling robots to competently perform everyday manipulation activities such as household chores exceeds, in terms of task,activity, behavior and context complexity, anything that we have so far investigated in motion planning, cognitive robotics, autonomous robot control and artificial intelligence at large. For achieving robust, adaptive, effective and natural performance of everyday manipulation tasks, it is not feasible to expect that programmers can equip the robots with plan libraries that cover such open-ended task spectrum competently.RoboHow.Cog targets at enabling autonomous robots to perform expanding sets of human-scale everyday manipulation tasks - both in human working and living environments. To this end, RoboHow.Cog will investigate a knowledge-enabled and plan-based approach to robot programming and control where knowledge for accomplishing everyday manipulation tasks is semi-automatically acquired from instructions in the World Wide Web, from human instruction and demonstration (videos), and from haptic demonstration.The knowledge-enabled control will be made possible through extensions of constraint- and optimization-based movement specification and execution methods that allow for the force adaptive control of movements to achieve the desired effects and avoid the unwanted ones. In addition, novel perception mechanisms satisfying the knowledge preconditions of plans and monitoring the effects of actions will make the RoboHow.Cog approach feasible.The software components that will come out of RoboHow.Cog will be integrated into complete generic robot control systems such as ROS, and, in particular, into Aldebaran's humanoid platform Romeo. RoboHow.Cog will strive to make the code of many of its components - and even of large parts of the Milestone demonstrations -- publicly available under free/open source software licenses.